论文部分内容阅读
为了降低单分类器财务困境预测的不确定性和不稳定性,本文通过多分类器组合来提高预测效果,提出了企业财务困境预测的多分类器混合组合模型,实现了并联组合和串联组合的优势互补作用.采用差异性原则和个体优化原则作为选择基本分类器的标准,在定义单类择优算子的基础上,设计了构建混合组合基本模块的流程算法以及混合组合模型内部并联结构的动态赋权机制和加权多数投票机制.以中国上市公司为对象的实证研究证实了该模型在提高平均预测准确率的同时大大降低了离散程度,统计分析表明:该模型显著优于现有的单分类器财务困境预测模型.
In order to reduce the uncertainty and instability of single classifier financial predicament prediction, this paper improves the forecasting effect through multi-classifier combination and proposes a multi-classifier hybrid combination model to predict the financial distress of enterprises, and achieves the parallel combination and the serial combination Advantages and Complements.Using the principle of difference and individual optimization as the standard to choose the basic classifier, based on the definition of a single class of preferred operator, the flow algorithm for building a mixed basic module and the dynamic structure of the parallel structure within the mixed model Empowerment Mechanism and Weighted Majority Voting Mechanism.An empirical study of Chinese listed companies demonstrates that this model greatly reduces the degree of dispersion while improving the accuracy of average prediction. Statistical analysis shows that this model is significantly superior to the existing single-class Financial Distress Prediction Model.